Estimated Costs Associated With Radiation Therapy for Positive Surgical Margins During Radical Prostatectomy

IMPORTANCE Prostate cancer is the most common malignant neoplasm among men and is the one with the highest positive surgical margin (PSM) rate. This high rate is due to the difficulty in balancing the risk of extraprostatic disease and excising periprostatic structures, which ultimately affects patients’ quality of life. In the case of a PSM, the appropriateness of adjuvant radiation therapy (aRT) should be discussed. The financial burden of PSMs on health systems has not been investigated. OBJECTIVE To estimate the financial costs associated with a PSM during radical prostatectomy on the basis of the odds of undergoing aRT. DESIGN, SETTING, AND PARTICIPANTS This cohort study used data on men with prostate cancer from the US National Cancer Database (January 1, 2010, through December 31, 2015). Data were requested in March 2019, accessed in April 2019, and analyzed in August 2019. EXPOSURE Treatment with radical prostatectomy followed by aRT, if indicated. MAIN OUTCOMES AND MEASURES The attributable risk fraction of PSMs on undergoing aRT was estimated from a logistic regression with aRT administration as the outcome. The analysis was adjusted for patients’ socioeconomic and demographic characteristics and tumor characteristics. The aRT cost for the year 2019 was calculated using the Medicare Physician Fee Schedule and the Hospital Outpatient Prospective Payment System. The fraction of this cost attributable to a PSM was estimated according to its attributable risk fraction. RESULTS In total, 230 175 men were identified (median [interquartile range] age at diagnosis, 62.0 [56.0-67.0] years). Overall, 22.8% of the patients had a PSM. Patients with PSMs were more likely than those without PSMs to be older (median [interquartile range] age, 62.0 [56.0-66.0] years vs 62.0 [57.0-67.0] years) and nonwhite (9320 patients [17.8%] vs 29 872 patients [16.8%]), to have higher comorbidity scores (1604 patients [3.1%] vs 4884 patients [2.7%] with a Charlson-Deyo Comorbidity Index score 2) and worse tumor characteristics (category T3 and T4 disease, 26 394 patients [50.3%] vs 36 040 patients [20.3%]), and to have lower socioeconomic indicators (median annual income <$30 000, 5708 patients [10.9%] vs 17 874 patients [10.1%]; proportion of individuals without a high school degree in the area 29%, 6925 patients [13.2%] vs 22 648 patients [12.7%]). In addition, PSMs were documented more frequently at nonacademic institutions than academic ones (31 702 patients [60.5%] vs 20 714 patients [39.5%]). A total of 11 585 patients (5.0%) underwent aRT, and 7698 of them (3.3%) had a PSM at the final pathology examination. When controlling for patients’ socioeconomic and demographic characteristics and tumor characteristics, men with PSMs were more likely than those with negative margins to undergo aRT, with an odds ratio of 3.79 (95% CI, 3.63-3.96; P < .001). The attributable risk fraction of the presence of a PSM on aRT was 44% (95% CI, 42%-45%). The attributable cost of a PSM was calculated as (continued) Key Points Question What are the costs associated with positive surgical margins (PSMs) during radical prostatectomy? Findings In this cohort study of 230 175 US men, the attributable cost of a PSM was $17 356. The overall health burden attributable to PSMs was estimated to be $52 068 000 each year. Meaning The financial burden of PSMs is substantial, and efforts in reducing the rate of PSMs could be associated with a reduction in the overall health costs associated with surgically treated prostate cancer. + Invited Commentary Author affiliations and article information are listed at the end of this article. Open Access. This is an open access article distributed under the terms of the CC-BY License. JAMA Network Open. 2020;3(3):e201913. doi:10.1001/jamanetworkopen.2020.1913 (Reprinted) March 31, 2020 1/11 Abstract (continued)continued) $17 356 (95% CI, $16 567-$17 751). Assuming 60 000 prostatectomies in 2019 and similar trends of PSM and aRT, the overall health burden attributable to PSMs was calculated to be $52 068 000 (95% CI, $49 701 000-$53 253 000). CONCLUSIONS AND RELEVANCE The estimated aRT cost attributable to the presence of a PSM was $17 356, resulting in $52 068 000 in spending on aRT in 2019. Strategies to reduce PSMs could be associated with a reduction in the overall health costs of surgically treated PCa. JAMA Network Open. 2020;3(3):e201913. doi:10.1001/jamanetworkopen.2020.1913


Introduction
One of the goals of the surgical treatment for prostate cancer (PCa) along with oncological control is to provide patients with an optimal quality of life. 1 In an effort to achieve that, surgeons must attempt a conservative dissection, allowing for maximal preservation of the structures surrounding the prostate, including the neurovascular bundles. The extent of the dissection will ultimately influence urinary continence and erectile function.
Considering the fact that surgeons walk on a fine line, trying to balance the risk of extraprostatic disease and the risk of resecting through tumorous tissue, the frequency of positive surgical margins (PSMs) is higher in men with prostate cancer compared with those with other malignant neoplasms. 2 Also, compared with the most common malignant neoplasm among women (ie, breast cancer), the PSM rate for surgically treated PCa is almost 4 times higher. 2 Because approximately 1 of 7 men will receive a diagnosis of PCa in their lifetime, the prognostic and financial implications of a PSM are obvious. 2 The PSM rate ranges from 10% to 25% and has been slightly fluctuating over time. 3,4 The introduction of robotic surgery initially led to an increase in the rate of PSMs, which later decreased in the postdissemination era. Recently, the use of preoperative multiparametric magnetic resonance imaging has been linked to a reduction in the PSM rate, 5 but thus far no randomized study has validated this observation, to our knowledge.
The National Comprehensive Cancer Network and American Urological Association guidelines suggest discussing the administration of adjuvant radiation therapy (aRT) for patients with PSMs, because the presence of this single factor has the highest association with recurrence in this population. 6,7 The European Association of Urology guidelines recommend the administration of aRT to individuals with a documented PSM in the case of locally advanced PCa. 8 Nevertheless, aRT is still underused, 9 and retrospective data suggest that surveillance followed by salvage radiation therapy in cases of recurrence may be a valid alternative. 10 Although the association of a PSM with a patient's prognosis is heterogeneous and depends on tumor category and grade, 11 it is evident that this occurrence has a heavy burden on the health care system. In this study, we aimed to estimate the cost of a PSM by using the US National Cancer Database (NCDB) to calculate the attributable risk fraction (ARF) of a PSM on aRT while incorporating 2019 reimbursement data.

Methods
of cancer. Details regarding the NCDB have been previously reported. 12 We queried the PCa NCDB for patients who underwent radical prostatectomy and had complete data on clinical, socioeconomic, demographic, pathology, and treatment covariates.
These data encompass deidentified information. The institutional review board of Icahn School of Medicine at Mount Sinai approved their use for the purpose of the study and waived the need for informed consent. This study follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. 13

Variables Definition
We considered patient socioeconomic characteristics, demographic factors, and tumor characteristics. Demographic variables included age, race, and Charlson-Deyo Comorbidity Index score. Socioeconomic factors included individual and neighborhood-level factors, including type of facility, primary payer, median annual income, proportion of individuals without a high school degree in the area, and distance from the facility. 14 Tumor characteristics included year of diagnosis, tumor category, Gleason score, lymph node invasion, and surgical margins status. On pathology, neoplastic tissue in contact with the inked margin denoted a PSM. Adjuvant radiation therapy was defined as the beginning of radiotherapy within 6 months of surgery.

Statistical Analysis
Descriptive statistics were generated; frequencies and proportions were reported for categorical variables, and medians and interquartile ranges were reported for continuous variables. Differences between medians and across frequencies were evaluated with the Kruskal-Wallis and the χ 2 test, respectively.
Our statistical analysis sensu stricto consisted of 2 main steps. First, we estimated how much the presence of a PSM at final pathology was associated with the subsequent administration of aRT.
Second, using the same regression model, we estimated the percentage and associated 95% CIs of how much of the administration of aRT can be attributed to the presence of a PSM.
For the first purpose, a multivariable logistic regression was fitted with aRT as the outcome. The role of PSMs in determining the administration of adjuvant treatments was investigated after adjusting for patient socioeconomic and demographic factors and tumor characteristics. Thereafter, the ARF of a PSM on the subsequent odds of undergoing aRT was calculated. This estimate (the ARF) reflects how much of the treatment (in this case aRT) can be attributed to a single variable, in our case the presence of a positive resection margin. 15 Statistical analyses were performed using Stata statistical software version 14 (StataCorp). All tests were 2-sided with a significance level set to P < .05.

Cost Estimation
In an effort to provide an estimate regarding the cost of the presence of a PSM, we considered the reimbursement rates for the year 2019 in the US in accordance with the Hospital Outpatient Prospective Payment System and the National Payment Amount for the Medicare Physician Fee Schedule. We considered the reimbursement for a conventional intensity-modulated radiation therapy as aRT for patients with PCa (ie, 70.2 Gy administered in 39 fractions). This was chosen because intensity-modulated radiation therapy is the most widely adopted treatment, while also compensating for the higher cost of proton treatment and lower cost of external beam radiation therapy. 16 The total amount of the reimbursement was multiplied by the ARF of the presence of a PSM in determining the administration of aRT. This provided us an estimate of how much of the cost of aRT can be attributed to a PSM.

Results
Overall, 230 175 men with complete data were identified. Patients received a diagnosis of PCa    (Table 2), and 2119 of them (or 1 of 13) underwent aRT (Table 1).
We estimated that the reimbursement for intensity-modulated radiation therapy treatment was $39 446 for the year 2019. Thus, the attributable cost of a PSM estimated on the odds of receiving aRT was $17 356 (95% CI, $16 567-$17 751). This was obtained by multiplying the ARF and its associated 95% CI (0.44; 95% CI, 0.42-0.45) by the cost of intensity-modulated radiation therapy ($39 446).
In accordance with the national trend of prostatectomies per year from the national inpatient sample, we estimated that approximately 60 000 radical prostatectomies were performed in 2019.
If 5% of the patients underwent aRT, the overall health burden attributable to PSMs would be  Year of diagnosis, median be attributed to PSMs in organ-confined disease. This was obtained by multiplying overall health burden attributable to PSMs ($52 068 000 and associated 95% CI) by the rate of aRT in case of organ-confined disease (18%).

Discussion
Because of the advent of robotic surgery, the rate of PSMs has been slightly decreasing over time; however, the overall PSM rate in the NCDB is still nonnegligible, with approximately 1 of 4 or 5 men who has a PSM after radical prostatectomy. 17 Such PSMs translate into worse prognosis 11,18,19 in terms of recurrence that is ultimately reflected in enormous health costs.
In this study, we have tried to estimate how much of the cost of aRT can be ascribed to the presence of PSMs on final pathology. To do so, we tried to predict the administration of aRT as the outcome of our multivariable analysis. Then, using the same model, we went backward in the prediction process and used the ARF of a variable (ie, PSM), to abstract how much of the outcome can be attributed to the same variable. After we obtained the ARF and associated 95% CIs, we multiplied this percentage by the cost of aRT to provide an estimate of the cost of a PSM.
In an effort to provide an estimate of how much of the cost of aRT can be attributed to PSMs, our primary analysis focused on the entire patient population who underwent radical prostatectomy between 2010 and 2015. Our model demonstrates that after adjusting for potential confounders, the presence of a PSM accounts for 44% of the decision to administer aRT following surgery. Of the cost of aRT, $17 356 could be attributed to the presence of a PSM on final pathology.

American Urological Association and National Comprehensive Cancer Network guidelines
suggest discussing the appropriateness of the administration of aRT in patients with a PSM. 3,4 However, the European Association of Urology guidelines suggest doing so in the case of non-organconfined disease, 5 and the present data seems to suggest that this is the preference among US urologists as well. In fact, it appears that approximately 1 of 13 men with a PSM in the context of organ-confined disease underwent aRT. Theoretically, these costs could be reduced with the implementation of well-structured training. In fact, although a PSM in the context of a non-organconfined disease is sometimes inevitable, a PSM in organ-confined disease is the result of a breach in the prostatic pseudocapsule in the area where the tumor is located and is more representative of a surgeon's mistake. One can argue that patients with non-organ-confined disease should undergo radiation therapy up front, and this would decrease the cost originating from a multimodal approach.  20 showed that the percentage of reduction in PSMs was greater in cases of non-organ-confined disease (approximately 25%), vs organ-confined disease (approximately 10%). A more structured training with the implementation of a specific curriculum could allow for potential PSM reduction and subsequent health care cost reduction. 21 Although it is true that the introduction of robotic surgery has allowed for the reduction of the overall rate of PSMs, our findings suggest that this occurrence is still quite frequent even in the postdissemination era.
The rate of PSMs is sometimes considered as a proxy of the quality of the surgery performed and, ultimately, of a surgical center. 22 We found that PSMs were more frequent in nonacademic centers and in older, nonwhite patients with more comorbidities and lower socioeconomic indicators, such as primary payer, median income, proportion of individuals without high school degree in the area, and distance from the facility. When taking all these factors into account, PSMs might also be a sign of health disparities that may warrant further research.
Our study, in keeping with previous findings, suggests that aRT is still underused. It is true that not all PSMs are equal and not all of them have the same prognostic outcome. Several studies have tried to address this point, aiming to define margins' features that have greater implications in terms of recurrence. Thus far, the Gleason score in the margin location 23,24 and a PSM length greater than 3 mm have been associated with a greater risk of biochemical recurrence, 18,21 whereas multiple margins or a single margin greater than 3 mm have been associated with a greater risk of metastasis in cases of non-organ-confined disease. 11 However, these data originate from analyses of retrospective studies and, to our knowledge, no prospective series has yet confirmed these results. The lack of available data describing margin features precluded us from taking this factor into account in our analysis.

Limitations
This study has some limitations. Despite revealing a tremendous financial burden for health care systems, this analysis represents an underestimation of the true cost of PSMs in the real-world setting. In fact, PSMs are well known to be associated with biochemical recurrence. 18,21,25 In addition, in the case of adverse pathology, PSMs are associated with a higher risk of developing metastasis during long-term follow-up. 26 Our estimates do not capture the subset of patients who have been offered surveillance and subsequently received salvage treatments (either salvage radiation therapy and/or salvage androgen deprivation therapy) because those data are not collected in the NCDB.
These factors render our estimate an underestimation of the true costs. In addition, the fact that the NCDB only captures approximately 70% of the newly diagnosed cases of PCa every year might influence our estimates. 12

Conclusions
To our knowledge, this study is the first to provide an estimate of the cost of a PSM. We estimated the aRT cost attributable to the presence of a PSM to be $17 356, resulting in $52 068 000 in spending on aRT in 2019. Even though these numbers represent an underestimation of the real PSM burden, they point to the fact that strategies to reduce PSMs could be associated with a reduction in the overall health costs of surgically treated PCa.

JAMA Network Open | Urology
Costs Associated With Radiation Therapy for Positive Surgical Margins During Radical Prostatectomy